Presentation is loading. Please wait.

Presentation is loading. Please wait.

® Hosted and Sponsored by Gold Sponsor Copyright © 2012Open Geospatial Consortium WaterMLWaterML 2.0 Time Series and NetCDF WaterML 78th OGC Technical.

Similar presentations


Presentation on theme: "® Hosted and Sponsored by Gold Sponsor Copyright © 2012Open Geospatial Consortium WaterMLWaterML 2.0 Time Series and NetCDF WaterML 78th OGC Technical."— Presentation transcript:

1 ® Hosted and Sponsored by Gold Sponsor Copyright © 2012Open Geospatial Consortium WaterMLWaterML 2.0 Time Series and NetCDF WaterML 78th OGC Technical Committee Austin, Texas (USA) Doug Palmer March 21, 2012

2 OGC ® Why? Why? Why? Lots of existing software can read NetCDF –With suitable conventions Archival format –XML is good as an exchange format, but a bit verbose –Metadata catalogues, such as THREDDS –Access to large data sets via OPeNDAP Copyright © 2012 Open Geospatial Consortium

3 OGC ® Use Cases Archival Storage –Need complete representation of WaterML 2.0 instance –Discoverable, comprehensible –Able to reconstruct WaterML 2.0 from encoding Data Payload –Timeseries result referenced by an observation Need to ensure that it remains comprehensible when separated from context –CSML Presenting Existing NetCDF as WaterML 2.0 –Import timeseries stored in NetCDF and present as WaterML 2.0 Import/Export Format for Tools and Software –Import hydrology data into visualisation, analysis, processing tools

4 OGC ® Data Models WaterML 2.0 –UML model –Metadata part of the model –Semantics provided by model and linked data –Complex, deep, flexible NetCDF –Multidimensional array-based model –Metadata stored in named attributes –Semantics provided by attribute and data conventions –Classic Data Model NetCDF-3 Simple, shallow, accessible (even by FORTRAN) Most conventions use the classic model Use this model as much as possible –Extended Data Model NetCDF-4 More complex, deep, flexible No common conventions that use the extensions Tools use the binary encodings but not the extensions to the data model Copyright © 2012 Open Geospatial Consortium

5 OGC ® NetCDF Conventions NetCDF Attribute Conventions –Useful everywhere –Require unit alignment (UCUM, UDUNITS) Attribute Conventions for Dataset Discovery (ACDD) –Useful for dataset catalogues –A good place to put some of the ISO (and other) metadata that appears in WaterML (thanks NASA GEO-IDE!) Climate and Forecast Conventions (CF) –Common –Discrete Sampling Geometries (Timeseries) –Location/Coordinate Reference System Conventions –Metadata and Ancillary Variable Conventions –But Not Stuff which is hard to map onto hydrology domain Standard names In future, feature types Copyright © 2012 Open Geospatial Consortium

6 OGC ® Encoding Strategies Copyright © 2012 Open Geospatial Consortium

7 OGC ® Gracefulness Copyright © 2012 Open Geospatial Consortium Completeness Compatibility Embedded XML Metadata Extended NetCDF Data Model Linked Metadata Dictionaries WaterML/O&M-specific Metadata Data Point Metadata Aggregate Cell Representation Location/CRS Conventions Discrete Sampling Geometries Conventions Ancillary Variables Conventions ACDD Conventions NetCDF Attribute Conventions Unit Conventions Archival Format Data Payload Import/Export NetCDF as WaterML 2.0 Works as a NetCDF Document

8 OGC ® Timeseries CF Discrete Sampling Geometries (Timeseries) Maps onto collections of timeseries from WaterML 2.0 monitoring points collecting simple, single property values –timeSeries feature type –Each monitoring point is a station –Independent timestamps for samples –X- and Y-coordinates latitude and longitude of monitoring points –Can map gridding onto coordinate reference system (CRS) Good correspondence to what is well-defined in the WaterML 2.0 specification Copyright © 2012 Open Geospatial Consortium

9 OGC ® Timeseries Example Copyright © 2011 Open Geospatial Consortium variables: double lat; lat:standard_name = “latitude”; lat:axis = “X”; double long; long:standard_name = “longitude”; long:axis = “Y”; int time(station, time); time:standard_name = “time”; time:units = “seconds since :26:00 +10:00”; time:axis = “T”; char station_id(station, station_name); station_id:standard_name = “station_name”; station_id:cf_role = “timeseries_id”; double stage(station, time); stage:standard_name = “stage”; data: lat = ; long = ; time = 0, 300; station_id = “Lake George 1”; stage = 1.20, 1.15;

10 OGC ® Timeseries (Future Issues) WaterML 2.0 is not conceptually limited to point timeseries –Transects –Swaths –Cross-sections –… CF timeseries feature types do not map well onto possible hydrological samples –horizontal transect vs. vertical profile –Possibility of multiple CRS series of monitoring points located via GPS/WGS84 sampling grids laid out via MGA94 More research needed... Copyright © 2011 Open Geospatial Consortium

11 OGC ® Dataset Metadata WaterML 2.0 has a lot of complex metadata –ISO –O&M procedure, times, bounding box, etc. –Timeseries descriptions Map useful bits onto NetCDF global attributes –Partial mapping of WaterML 2.0 metadata onto NetCDF attributes Attribute Conventions for Dataset Discovery (ACDD) Standard NetCDF attribute conventions –Additional attributes for domain-specific information Procedures Gauge datum Vocabularies … Permit a complete description for reconstruction –An attribute with embedded XML –I feel unclean! Copyright © 2012 Open Geospatial Consortium

12 OGC ® Data Point Metadata Each data point can have associated metadata –Quality, Nil reason, Censored reason, Interpolation type, Processing, Comment, Related observation, Accuracy, Qualifier, Unit WaterML 2.0 timerseries have default values –Map defaults onto variable attributes Map useful per-point metadata onto CF-style ancillary variables –Include what is fit for purpose –Some fields require use of the enhanced model Comments –Some fields cannot be mapped onto a simple ancillary variable Arbitrary qualifiers Use XML representation, if needed Copyright © 2012 Open Geospatial Consortium

13 OGC ® Restrictions Restrict ourselves to situations that can be sensibly represented as NetCDF –Single observations –Collections of monitoring points collecting homogeneous data –Monitoring points collecting multi-property samples –Simple aggregation Can represent most WaterML 2.0 –But beware expanding complexity –Beware crossing the “classic”/”enhanced” NetCDF border Copyright © 2012 Open Geospatial Consortium

14 OGC ® Mapping and Encoding Strategies Copyright © 2012 Open Geospatial Consortium

15 OGC ® Terms and Vocabularies WaterML 2.0 tends to use URI-like terms –http://www.opengis.net/def/timeseriesType/WaterML/2.0/MinSucchttp://www.opengis.net/def/timeseriesType/WaterML/2.0/MinSucc –http://sweet.jpl.nasa.gov/2.3/stateVisibility.owl#Sunnyhttp://sweet.jpl.nasa.gov/2.3/stateVisibility.owl#Sunny – Poor –Definitions provided by resolving the URI NetCDF tends to use token-like terms –quality_good, outside_valid_range –Definitions provided by a conventions document Copyright © 2012 Open Geospatial Consortium

16 OGC ® Embedding Vocabularies Leverage CF Flags Conventions –Tokenise terms –Construct an enumeration of terms Copyright © 2012 Open Geospatial Consortium... /terms/ Poor... byte measure_quality(station, time);... measure_quality:flag_masks = 7b; measure_quality:flag_values = 1b, 2b, 3b, 4b; measure_quality:flag_meanings = “good poor biased invalid”; measure_quality = 2b,...;

17 OGC ® Embedding Vocabularies Not finished yet … –Map tokens onto vocabulary –Link the variable to the vocabulary Can be an external link –Link the variable to the type of vocabulary Simple embedded dictionary SKOS concept scheme GML dictionary … more Copyright © 2012 Open Geospatial Consortium :measure_quality_dictionary = “good=http://www.example.com/terms/2.3#Good...”; byte measure_quality(station, time);... measure_quality:flag_vocabulary = “measure_quality_dictionary”; measure_quality:flag_vocabulary_typeof = “simple”;

18 OGC ® Linked (Meta)Data Lots of WaterML 2.0 information is via linked data –E.g. feature of interest, procedure –Use of Xlink –xlink:title and xlink:href are expected Use link attributes –Name/link pattern Drawn from unofficial metadata_link attribute in ACDD Copyright © 2012 Open Geospatial Consortium :sampled_feature = “Lake George”; :sampled_feature_link = “http://www.example.com/foi/LG”;

19 OGC ® Aggregate Data Many WaterML 2.0 timeseries represent aggregates –Tipping bucket –Maximum temperature CF Conventions allow cell boundaries and methods –In the time domain Copyright © 2012 Open Geospatial Consortium variables: double maximum_stage(station, time); maximum_stage:cell_methods = “time: maximum”; int time(station, time); time:units = “seconds since :27:00 +10:00”; time:bounds = “time_bounds”; int time_bounds(station, time, bounds); data: time = 0, 60, 180, 240; time_bounds = -60, 0, 0, 60, 60, 180, 180, 240; maximum_stage = 3.0, 3.5, 3.4, 3.3;

20 OGC ® Summary Mapping between the WaterML 2.0 model and NetCDF –Not complete But covers the major cases Some difficulties with both the model and the encoding –Graceful complexity/completeness/compatibility trade-off Useful NetCDF documents Bridge between NetCDF and WaterML 2.0 vocabularies –Dictionary construction Discussion Paper –OGC WaterML 2.0 – Timeseries – NetCDF Discussion Paper Copyright © 2012 Open Geospatial Consortium


Download ppt "® Hosted and Sponsored by Gold Sponsor Copyright © 2012Open Geospatial Consortium WaterMLWaterML 2.0 Time Series and NetCDF WaterML 78th OGC Technical."

Similar presentations


Ads by Google